ARTICLE IN PRESS

THE JOURNAL OF PEDIATRICS • www.jpeds.com ORIGINAL

ARTICLES
Increased Fracture Risk with Furosemide Use in Children with
Congenital Heart Disease
Ji Haeng Heo, PhD1, Karen L. Rascati, PhD2, Keila N. Lopez, MD3,4, and Brady S. Moffett, PharmD, MPH3,4

Objectives To determine the association of furosemide therapy with the incidence of bone fractures in children
with congenital heart disease.
Study design We conducted a retrospective cohort study with data extracted from the 2008-2014 Texas Med-
icaid databases. Pediatric patients aged <12 years diagnosed with congenital heart disease, cardiomyopathy, or
heart failure were included. Patients taking furosemide were categorized into a furosemide-adherent group (medi-
cation possession ratio of ≥70%), and a furosemide-nonadherent group (medication possession ratio of <70%). A
third group of patients was matched to the furosemide user groups by using propensity score matching. A multi-
variate logistic regression and Cox proportional hazard model with a Kaplan–Meier plot (time-to-fracture) were used
to compare the 3 groups, controlling for baseline demographics and clinical characteristics.
Results After matching, 3912 patients (furosemide adherent, n = 254; furosemide nonadherent, n = 724; no fu-
rosemide, n = 2934) were identified. The incidence of fractures was highest for the furosemide-adherent group (9.1%;
23 of 254), followed by the furosemide-nonadherent group (7.2%; 52 of 724), which were both higher than for pa-
tients who did not receive furosemide (5.0%; 148 of 2934) (P < .001). Using logistic regression, both furosemide
groups were more likely to have fractures than the no furosemide group: furosemide-adherent OR of 1.9 (95% CI,
1.17-2.98; P = .009); furosemide nonadherent OR of 1.5 (95% CI, 1.10-2.14; P = .01). In the Cox proportional hazard
model, the risk of fractures for the furosemide-adherent group was significantly higher compared with the no fu-
rosemide group (HR, 1.6; 95% CI, 1.00-2.42; P = .04).
Conclusions Furosemide therapy, even with nonconsistent dosing, was associated with an increased risk of bone
fractures in children with congenital heart disease. (J Pediatr 2018;■■:■■-■■).

F
urosemide, a potent diuretic, increases the urinary loss of potassium, calcium, and magnesium by inhibiting the passive
reabsorption of these ions in the loop of Henle of the kidney.1,2 Furosemide can thus cause hypercalciuria and
nephrocalcinosis.3 The excretion of calcium in the urine may cause a loss of bone mineral density, which can lead to
osteoporosis.4-6 An observational study indicated that for the adult population studied, any use of loop diuretics was associ-
ated with an increased risk of any fracture (crude 51% [odds ration (OR), 1.51; 95% CI, 1.48-1.55]; adjusted 4% [OR, 1.04;
CI, 1.01-1.07]).7 In a meta-analysis that studied the association between loop diuretic uses and the risk of fractures for adult
patients, compared with nonloop diuretic users, loop diuretics users had an approximately 15% higher risk of total fractures.8
Furosemide is less commonly prescribed in children than adults. Specific children, including those with congenital heart defects
(CHD), have a higher use of this medication.1 More specifically, in children who have CHD and cardiomyopathies, diuretics
often are given to treat symptoms of heart failure.9 However, given studies conducted in adults with furosemide use, concerns
remain about loop diuretics, particularly with chronic use, for children. Furosemide is specifically related to higher rates of
hypercalciuria.10 Atkinson et al reported that treatment with any type of diuretic in infants was associated with an abnormal
renal loss of calcium, sodium, chloride, and potassium and others have cautioned that children with CHD or who have car-
diomyopathy and who are prescribed loop diuretics may be at particular risk for developing metabolic bone disease.11
The purpose of this study was to determine the association of furosemide therapy with fractures in children with CHD.

Methods
Data were extracted from the Texas Medicaid Database, which consists of insur-
ance claims that include demographic, medical, and prescription claims between
January 1, 2008 and December 31, 2014. The Texas Medicaid program provides
From the 1Genesis Research, Hoboken, NJ; 2Health
Outcomes and Pharmacy Practice, College of Pharmacy,
The University of Texas at Austin, Austin; 3Department of
Pediatrics, Baylor College of Medicine, Houston; and
4Department of Pharmacy, Texas Children’s Hospital,

Houston, TX
The authors declare no conflicts of interest.
CHD Congenital heart defect Portions of this study were presented at the American
HR Hazard ratio College of Cardiology Scientific Sessions, March 17-19,
2017, Washington, DC.
ICD-9-CM International Classification of Diseases, Ninth Revision, Clinical Modification
MPR Medication possession ratio 0022-3476/$ - see front matter. © 2018 Elsevier Inc. All rights
PPIs Proton pump inhibitors reserved.
https://doi.org10.1016/j.jpeds.2018.03.077

FLA 5.5.0 DTD ■ YMPD9926_proof ■ May 9, 2018

THE JOURNAL OF PEDIATRICS • www.jpeds.com
health insurance coverage for low-income families, individu-
als with chronic disabilities, blind persons, low-income preg-
nant women, elderly people or seniors, nondisabled children,
and caretakers of dependent children. Medicaid enrollment in
the state of Texas for clients <21 years of age is approxi-
mately 3 million for 2016.12
The study was approved by the University of Texas at Austin
Institutional Review Board and by the Texas Health and Human
Services Commission. The following information was ex-
tracted from the Texas Medicaid database: date of birth, sex,
race/ethnicity, start and end dates of health plan enrollment,
International Classification of Diseases, Ninth Revision, Clini-
cal Modification (ICD-9-CM) diagnostic codes, Healthcare
Common Procedure Coding System codes, service dates, quan-
tity of the medication dispensed, the number of days of supply,
National Drug Code, Generic Sequencing Number, and Ameri-
can Hospital Formulary Service number.
Study Design
A retrospective cohort study design was used to assess the as-
sociation of furosemide use and fractures. Both inpatients and
outpatients were included in the study if they were <12 years
of age and had ≥1 claim with a diagnosis of CHD, cardiomy-
opathy, or heart failure. Many patients were diagnosed as infants;
therefore, 1 year of previous use before the index date was not
required, but all were required to have ≥1-year of follow-up
data (ie, be enrolled and using services) past their medica-
tion index date. Patients were excluded if the first date for a
diagnosis claim for CHD, cardiomyopathy, or heart failure oc-
curred after the medication index date. In addition, if pa-
tients had any claims for diuretic prescriptions or ICD-9-
cardiomyopathy codes for fractures on or before their medi-
cation index date they were also excluded from analysis. The
study timeframe was described in Appendix 1 (available at
www.jpeds.com).
Under the assumption that a patient used furosemide chroni-
cally, those who met study criteria were divided into 3 groups.
Furosemide-Adherent Group. The date of the first prescrip-
tion for furosemide was considered the patient’s medication
index date. If the patient had at least 256 days of furosemide
prescriptions during the first year post-index (ie, medication
possession ratio [MPR] of ≥ 70%), they were categorized into
the furosemide-adherent group. A wide range of cutoff ad-
herence values for the MPR (63%-89%) have been used in pre-
vious adherence studies.13 However, the MPR cutoff of 70%
was used for this study based on the distribution of adher-
ence in this sample (ie, natural break). A sensitivity analysis
was conducted to validate the base model using the more com-
monly used 80% cutoff MPR.
Furosemide-Nonadherent Group. Again, the date of the first
prescription for furosemide was considered the patient’s medi-
cation index date. If the patient had <256 days of furosemide
prescriptions during the first year postindex (ie, a MPR
of <70%), they were categorized into the furosemide-
nonadherent group.
2 Heo et al
FLA 5.5.0 DTD ■ YMPD9926_proof ■ May 9, 2018
Volume ■■ • ■■ 2018
No Furosemide Group. If patients did not have a furose-
mide prescription at any time, but had another diuretic pre-
scription, their index date was defined as the first date of this
diuretic prescription fill. The other type of diuretics in-
cluded potassium-sparing diuretics and thiazide diuretics. If
patients did not have any prescriptions for diuretics, a random
index date was generated from the list of medication fill dates
for that patient within one year from the first prescription claim
date. Patients with nonfurosemide diuretics or no diuretics were
combined to form the no furosemide group.
Study Outcomes and Covariates
The primary outcome was the new occurrence of a bone frac-
ture within the postindex period. To control for other factors
that might be associated with the incidence of fractures, ad-
ditional variables were included as covariates in the multi-
variate analyses. First, patients were divided into those who only
had a CHD diagnosis and those that had either a cardiomy-
opathy and/or heart failure diagnosis in addition to a CHD
diagnosis. Second, 2 diseases indicators were selected as
covariates to assess comorbidity: a diagnosis of bronchopul-
monary dysplasia (ICD-9-CM 770.7) and a diagnosis of low
birth weight or prematurity (ICD-9-CM 764.x, 765.x). Next,
clinical factors were included to adjust severity of disease: heart-
related surgery (eg, heart surgery, aortic valve repair) and use
of proton pump inhibitors (PPIs; ie, omeprazole, esomeprazole,
lansoprazole, rabeprazole, pantoprazole, and dexlansoprazole),
which have also been shown to increase fracture risk.14-16 In
addition, the use of less commonly used medications—H2-
antagonist, beta-blockers, and calcium or vitamin D
supplementation—were added in a sensitivity analysis. These
agents were identified only when their claims were found
between the first date of diagnosis of CHD, cardiomyopathy,
or heart failure and the first date of the occurrence of a frac-
ture. For those who did not have any fractures in the study
period, a random date within 1 year from the index date was
used instead of first fracture date. In addition, other demo-
graphic factors included in the model were age at index date;
sex; and race/ethnicity (white, black, Hispanic, and others
[Asian, Native Hawaiian, or uncategorized race]). The de-
tailed ICD-9-CM codes for CHD, cardiomyopathy, heart failure,
and fractures used in this study are described in Appendix 2
and Appendix 3 (available at www.jpeds.com).
Statistical Analyses
Baseline characteristics and treatment variables were com-
pared using c2 tests for all categorical variables and ANOVA
tests for all continuous variables. Matching was performed using
the propensity score matching method to reduce the bias in
covariates among 3 groups. Propensity scores were generated
using logistic regression and the matching used a greedy al-
gorithm, which uses the nearest available pair matching
methods.17 Covariates used for logistic regression included all
covariates as described, including both demographic and clini-
cal factors. The incidence of fracture occurrence was com-
pared among groups using c2 tests. To estimate the odds of a
fracture in the study population, a logistic regression of the
■■ 2018 ORIGINAL ARTICLES

matched patients, controlling for covariates, was conducted.
In addition, the log-rank test was used to compare survival dis-
tribution among the 3 groups, and a Cox proportional hazard
model with a Kaplan-Meier plot (survival curve) was used to
assess time to fracture controlling for covariates. All data man-
agement and statistical analyses were completed using SAS ver-
sions 9.4 (SAS Institute Inc, Cary, North Carolina) and Stata
Statistical Software: Release 14 (StataCorp, College Station,
Texas). P < .05 was considered statistically significant a priori.
Results
A total of 53 725 patients (furosemide-adherent, n = 466; fu-
rosemide nonadherent, n = 2810; no furosemide, n = 50 449)
met the study criteria (Appendix 4; available at
www.jpeds.com). The median follow-up periods were about
3 years for each of the 3 groups. Significant differences in all
demographic and clinical characteristics (P < .001, except sex:
P < .02; Table I) were found between the unmatched groups
at baseline. After propensity score matching, a final sample size
of 3912 patients (furosemide-adherent, n = 254; furosemide
nonadherent, n = 724; no furosemide, n = 2934) was created
with no differences in all baseline demographics (Table I). The
vast majority (>91%) had ≥2 claims for CHD. Few patients
had any claim for nonfurosemide diuretics, such as thiazide
or potassium-sparing diuretics: furosemide-adherent (3.9%,
n = 10), furosemide nonadherent (1.5%, n = 11), and no fu-
rosemide group (1.7%, n = 51).
The incidence rate of a fracture was significantly higher for
the furosemide-adherent group compared with both the fu-
rosemide nonadherent group and the no furosemide group for
both the unmatched and matched cohorts (P < .001). Before
matching the fracture rate was highest for the furosemide-
adherent group (9.0%), followed by the furosemide
nonadherent group (5.6%), which were both higher than for
patients who did not receive furosemide (3.6%). After pro-
pensity score matching, the fracture rate remained highest for
the furosemide-adherent group (9.1%), followed by the
furosemide-nonadherent group (7.2%), which were both higher
than for patients who did not receive furosemide (5.0%).
Appendix 5 and Appendix 6 (available at www.jpeds.com)
show the frequencies of the site of fractures per each group
for original and matched population.
In the logistic regression model, the odds of having a frac-
ture were 87% higher for the furosemide-adherent group com-
pared with the no furosemide group, despite controlling for
covariates (OR, 1.9; 95% CI, 1.17-2.98; P = .009; Table II). The

Table I. Baseline characteristics of study population

Original cohort* (n = 53 725) Propensity score matched cohort† (n = 3912)
Furosemide Furosemide No Furosemide Furosemide No
n (%) adherent nonadherent furosemide P value‡ adherent nonadherent furosemide P value‡
Total patients 466 2810 50 449 254 724 2934
Age
Mean (SD)§ 1.29 (2.13) 0.85 (1.70) 1.65 (1.98) <.001 1.67 (2.30) 1.51 (2.06) 1.58 (2.08) .52
0 286 (61.4) 2008 (71.5) 20 237 (40.1) <.001 124 (48.8) 364 (50.3) 1398 (47.7) .73
1-2 85 (18.2) 404 (14.4) 16 722 (33.2) 65 (25.6) 183 (25.3) 754 (25.7)
3-11 95 (20.4) 398 (14.2) 13 490 (26.7) 65 (25.6) 177 (24.4) 782 (26.6)
Sex
Female 225 (48.3) 1297 (46.2) 24 662 (48.9) .02 117 (46.1) 350 (48.3) 1449 (49.39) .56
Male 241 (51.7) 1513 (53.8) 25 787 (51.1) 137 (53.9) 374 (51.7) 1485 (50.61)
Race/ethnicity
White 45 (9.7) 315 (11.2) 5405 (10.7) <.001 24 (9.5) 67 (9.2) 285 (9.7) .46
Black 23 (4.9) 158 (5.6) 3757 (7.5) 18 (7.1) 38 (5.3) 138 (4.7)
Hispanic 212 (45.5) 1281 (45.6) 29 512 (58.5) 123 (48.4) 393 (54.3) 1520 (51.8)
Others 186 (39.9) 1056 (37.6) 11 775 (23.3) 89 (35.0) 226 (31.2) 991 (33.8)
Only CHD¶
No 318 (68.2) 1490 (53.0) 2735 (5.4) <.001 109 (42.9) 318 (43.9) 1308 (44.6) .85
Yes 148 (31.8) 1320 (47.0) 47 714 (94.6) 145 (57.1) 406 (56.1) 1626 (55.4)
Bronchopulmonary dysplasia
No 354 (76.0) 2449 (87.2) 46 502 (92.2) <.001 203 (79.9) 587 (81.1) 2309 (78.7) .35
Yes 112 (24.0) 361 (12.8) 3947 (7.8) 51 (20.1) 137 (18.9) 625 (21.3)
Low birth weight/premature
No 242 (51.9) 1540 (54.8) 32 062 (63.6) <.001 145 (57.1) 411 (56.8) 1621 (55.3) .68
Yes 224 (48.1) 1270 (45.2) 18 387 (36.4) 109 (42.9) 313 (43.2) 1313 (44.7)
Heart-related surgery
No 396 (85.0) 2328 (82.9) 50 020 (99.2) <.001 238 (93.7) 671 (92.7) 2772 (94.5) .18
Yes 70 (15.0) 482 (17.1) 429 (0.8) 16 (6.3) 53 (7.3) 162 (5.5)
PPI
No 283 (60.7) 2083 (74.1) 46 684 (92.5) <.001 189 (74.4) 588 (81.2) 2306 (78.6) .06
Yes 183 (39.3) 727 (25.9) 3765 (7.5) 65 (25.6) 136 (18.8) 628 (21.4)

*Furosemide-adherent: ≥ 70% MPR first year after index, furosemide nonadherent: < 70% MPR.
†Propensity score matching using a greedy algorithm (nearest available pair matching methods).
‡c2 test.
§ANOVA.
¶Yes (only CHD) vs no (dual diagnosis CHD/cardiomyopathy or CHD/cardiomyopathy or only cardiomyopathy or only heart failure or comorbidity).

Increased Fracture Risk with Furosemide Use in Children with Congenital Heart Disease 3

FLA 5.5.0 DTD ■ YMPD9926_proof ■ May 9, 2018

THE JOURNAL OF PEDIATRICS • www.jpeds.com Volume ■■

Table II. Logistic regression of fractures by furosemide use controlling for covariates*
Categories OR 95% CI Wald statistic P value†
Furosemide use‡
No furosemide Reference
Furosemide nonadherent 1.537 1.104-2.139 0.430 .01
Furosemide adherent 1.868 1.172-2.977 0.625 .009
Age 1.150 1.076-1.229 0.140 <.001
Sex
Male Reference
Female 0.934 0.707-1.234 −.068 .63
Race/ethnicity
White Reference
Black 0.488 0.239-0.994 −0.718 .05
Hispanic 0.632 0.420-0.950 −0.459 .03
Others 0.599 0.382-0.938 −0.513 .03
Only CHD§
No Reference
Yes 0.603 0.447-0.815 −0.505 .001
Bronchopulmonary dysplasia
No Reference
Yes 1.314 0.888-1.946 0.273 .18
Low birth weight/premature
No Reference
Yes 1.190 0.843-1.679 0.174 .32
PPI
No Reference
Yes 1.586 1.146-2.195 0.462 .005
Heart-related surgery
No Reference
Yes 1.115 0.660-1.883 0.109 .68

*Likelihood Ratio c2: 64.60 (P < .001); model fit statistics (Akaike information criterion): 1,672.07; percent concordant: 64.8%.
†P < .05.
‡Nonadherent furosemide ≤70% MPR first year after index; adherent furosemide ≥70% MPR.
§Yes (only CHD) vs no (dual diagnosis CHD/cardiomyopathy or CHD/cardiomyopathy or only cardiomyopathy or only heart failure or comorbidity).

furosemide-nonadherent group also had a 54% higher odds
of having fractures than the no furosemide group (OR, 1.5;
95% CI, 1.10-2.14; P = .01). Significant covariates for frac-
tures were age, race/ethnicity, CHD only diagnosis, and PPI
use. As each year in age increases, the odds of a fracture in-
creased 1.2 times (OR, 1.15; 95% CI, 1.08-1.23; P < .001). His-
panic and black patients showed a reduced risk of fracture
compared with white patients (Hispanic: OR, 0.6 [95% CI, 0.42-
0.95; P = .03]; black: OR, 0.5 [95% CI, 0.24-0.99; P = .05]).
When patients had only a CHD diagnosis, the odds of frac-
tures were lower than patients with a comorbid diagnosis of
CHD and cardiomyopathy or heart failure (OR, 0.6; 95% CI,
0.45-0.82; P = .001). The OR for PPIs users was 1.6 versus PPIs
nonusers (95% CI, 1.15-2.20; P = .005).
Table III shows the results of the Cox proportional hazards
regression model comparing the occurrence of fractures among
cohorts including a time-to-fracture component while con-
trolling for covariates. The proportional hazards assumption
was met, meaning that the hazards ratio is a constant that
does not depend on time (P = .42). This model indicated
that furosemide-adherent patients had a 56% (HR, 1.6; 95%
CI, 1.00-2.42; P = .04) increased risk of fractures consistently
over 3 years, compared with furosemide nonusers. The
furosemide-nonadherent group also had an increased risk
versus the no furosemide group, but this increase trended
toward but did not reach statistical significance (HR, 1.37;
4 Heo et al
95% CI, 1.00-1.89; P = .05). Similar to the logistic regression,
significant covariates were race/ethnicity, CHD only, and PPI
use. Hispanic and black patients showed a reduced risk of
fracture compared with white patients (Hispanic: OR, 0.6
[95% CI, 0.44-0.94; P = .02]; black: OR, 0.5 [95% CI, 0.25-
0.98; P = .04]). Patients with only CHD had a 34% lower risk
of fractures compared with patients with dual diagnoses (ie,
CHD/cardiomyopathy or CHD/heart failure) (HR, 0.7; 95%
CI, 0.49-0.88; P = .004). PPIs users had 47% higher risk of
fracture than nonusers (HR, 1.5; 95% CI, 1.08-2.00; P = .02).
The log-rank test showed that furosemide users have signifi-
cantly different survival distributions compared with nonusers
(c2 = 8.17; P = .02). Furosemide-adherent patients had a shorter
time to fracture as demonstrated by Kaplan-Meier analysis
(Figure).
The 2 sensitivity analyses using results with smaller cell sizes:
using a cutoff MPR of 80%, and the addition of 3
covariates—H2-antagonists, beta-blocker use, and calcium or
vitamin D supplementation—showed the same direction of
results as the base model (Appendixes 7-10; available at
www.jpeds.com), and the regression analyses indicated a sig-
nificant difference for furosemide users. For the sensitivity
analyses applied to the Cox proportional hazards model, again
results were in the same direction, but no longer statistically
different at the alpha level of .05 (P = .08 for MPR 80% and
P = .16 when 3 new covariates were added).

FLA 5.5.0 DTD ■ YMPD9926_proof ■ May 9, 2018

■■ 2018 ORIGINAL ARTICLES

Table III. Cox proportional hazard model of fractures by furosemide use controlling for covariates*
Categories Hazard Ratio 95% CI* Coefficient P value†
Furosemide use‡
No furosemide Reference
Furosemide nonadherent 1.372 0.998-1.885 0.316 .05
Furosemide adherent 1.557 1.001-2.423 0.443 .04
Age 1.122 1.052-1.198 0.115 <.001
Sex
Male Reference
Female 0.940 0.719-1.229 −.062 .65
Race/ethnicity
White Reference
Black 0.495 0.250-0.981 −0.702 .04
Hispanic 0.641 0.435-0.944 −0.445 .02
Others 0.683 0.445-1.048 −0.381 .08
Only CHD§
No Reference
Yes 0.657 0.492-0.875 −0.421 .004
Bronchopulmonary dysplasia
No Reference
Yes 1.292 0.890-1.877 0.257 .18
Low birth weight/premature
No Reference
Yes 1.208 0.867-1.683 0.189 .26
PPI
No Reference
Yes 1.469 1.078-2.002 0.385 .02
Heart-related surgery
No Reference
Yes 1.039 0.630-1.715 .039 .88

*Likelihood ratio c2: 46.90 (P < .001); model fit statistics (Akaike information criterion): 3,543.52.
†P < .05.
‡Nonadherent furosemide ≤70% MPR first year after index; adherent furosemide ≥70% MPR.
§Yes (only CHD) vs no (dual diagnosis CHD/cardiomyopathy or CHD/cardiomyopathy or only cardiomyopathy or only heart failure or comorbidity).

Figure. Kaplan–Meier curve of risk of fractures by furosemide use after 3 years of follow-up. Log-rank test (furosemide-
adherent vs furosemide nonadherent: c2 = 0.59, P = .44; furosemide-adherent vs no furosemide: c2 = 5.47, P = .02; furose-
mide nonadherent vs no furosemide: c2 = 4.2, P = .04).

Increased Fracture Risk with Furosemide Use in Children with Congenital Heart Disease 5

FLA 5.5.0 DTD ■ YMPD9926_proof ■ May 9, 2018

THE JOURNAL OF PEDIATRICS • www.jpeds.com Volume ■■

disease.39-41 It is important to note that the effect of loop di-

Discussion
uretic therapy on fracture persisted when adjusted for PPI use.
Fractures may be able to be mitigated by decreased use of
We found that patients with CHD on chronic furosemide
PPIs.41-43 Clinicians should be cognizant of the increased risk
therapy were nearly twice as likely to have a fracture com-
of fracture when using PPIs in children with CHD on chronic
pared with patients who were not on this therapy. The role of
loop diuretic therapy and tailor use accordingly.
diuretics in fracture is further supported by the increased risk
The limitations of this report are those that are germane to
of fracture in those patients who had more prescription refills
large database analyses. The current analysis only allows the
for a diuretic, suggesting a dose–response type effect.
reporting of an association between diuretic use and fracture
Although the finding of diuretic-associated fracture is well-
incidence. We believe that a causal relationship is likely, based
known in adult patients, this result is unique in this high-
on the clinical pharmacology of loop diuretics and the prior
risk subset of the pediatric population.7,18,19 It is also unique
literature, but cannot establish causation. There is an assump-
that in our analysis, the patient population evaluated is older
tion made that patients who refilled medications were actually
than previous reports, demonstrating bone mineral density
taking those medications, which we also cannot validate. The
defects in premature neonates and infants.20-24 The use of fu-
claims database also limited the validation of sensitivity between
rosemide in pediatric patients with CHD is common, and our
prescribed and filled prescription because Texas Medicaid da-
findings may spur further investigation and potentially a higher
tabase included only filled prescription data. Claims databases
clinical suspicion to evaluate for fractures.25 Additionally, there
do not have the granularity to assess pertinent clinical infor-
were several other findings in our report which warrant dis-
mation, such as bone mineral density values, severity of heart
cussion and evaluation from a clinical perspective.
disease, vitamin D status, calcium/vitamin D supplementa-
We noted that patients with a dual diagnosis of CHD/
tion, or fracture etiology. We were unable to quantify immo-
cardiomyopathy or CHD/heart failure were at a greater risk
bility, infection, inflammation, or dietary calcium using the
of fracture than those patients with only CHD. Prior investi-
claims database. However, McNally et al reported that the pa-
gations have noted the risk of bone fracture in infants with
tients who receive a standard of care did not show additional
CHD admitted to the cardiac intensive care unit, although it
cases of hypercalcemia or hypercalciuria.44 We would not expect
is unclear as to the etiology for the fractures reported.21 Vitamin
that there would be a difference in over-the-counter calcium
D deficiency is known to be common in patients with CHD
or vitamin D supplementation between the high-dose and low-
and cardiopulmonary bypass has been shown to decrease
dose furosemide groups, or nonusers group. The sensitivity
vitamin D levels in patients undergoing surgery for CHD.26,27
analysis was conducted—controlling for calcium or vitamin
Additionally, patients who have undergone a single ventricle
D supplement—to reduce potential bias and showed the same
palliation with the Fontan procedure have demonstrated de-
direction with primary results. This study can serve as the basis
creases in bone mineral density at a median of 10 years after
for future investigations using robust clinical data. Next, this
the surgical procedure.28 However, vitamin D has also been
was a nonrandomized retrospective analysis; therefore, the pres-
shown to be low in pediatric patients with heart failure and
ence of differences in baseline patient characteristics can result
no CHD.29 The lower levels of vitamin D could be a risk factor
in bias results. Although propensity score matching and re-
for fracture in this patient population. Overall, patients with
gressions were used to account for some baseline differences,
CHD only were less likely to experience a fracture, which may
there may be unmeasured variables that can introduce bias.
be due to lower doses or frequency of furosemide when com-
Next, thiazide is known to decrease renal calcium excre-
pared with patients with heart failure (P < .001). The pres-
tion and have been associated with increased bone mineral
ence of CHD, and the underlying genotype that give rise to
density, resulting likely decreased the risk of fractures. 45
this phenotype, seem to influence the risk of fracture. Given
However, in our matched population, only 10 patients who took
these findings, vitamin D supplementation should be consid-
≥1 claim for thiazide were identified in furosemide group. More-
ered in children with CHD on chronic loop diuretic therapy.30
over, because a 1-time administration of thiazide would not
A difference in sex with regard to fracture risk was not seen,
be enough to decrease the chance of a fracture, the use of
despite data suggesting that this may play a role.31 White pa-
thiazides was not considered as a covariate. Sedative medica-
tients were at a greater risk of developing a fracture in com-
tions that may be related to fractures encompass a large number
parison with Hispanic or black patients in our analysis. The
of medication classes and were considered too broad to be a
racial/ethnic differences in bone mineral density and frac-
useful covariate. Overall, despite the limitations associated with
ture risk are well-documented.7,31-34 For example, it has been
using claims data, the data presented are similar to the results
reported that blacks with osteoporosis have higher bone density
of other investigations in adults and make a compelling case
scores as compared with white patients, which in turn is related
for further investigations into bone health in pediatric pa-
to a lower incidence rate of fractures.35 These factors should
tients with CHD who are treated with loop diuretics. ■
be taken into account when assessing a patient for chronic di-
uretic therapy.
Submitted for publication Nov 13, 2017; last revision received Mar 11, 2018;
The use of PPIs was associated with an increased risk of frac- accepted Mar 30, 2018
ture, a finding previously reported.36-38 Despite this risk, these Reprint requests: Ji Haeng Heo, PhD, Genesis Research, 5 Marine View
agents are frequently prescribed in children with heart Plaza, Hoboken, NJ 07030. E-mail: jihaeng@genesisrg.com

6 Heo et al

FLA 5.5.0 DTD ■ YMPD9926_proof ■ May 9, 2018

■■ 2018
References
1. Lim LS, Fink HA, Blackwell T, Taylor BC, Ensrud KE. Loop diuretic use
and rates of hip bone loss and risk of falls and fractures in older women.
J Am Geriatr Soc 2009;57:855-62.
2. Ho KM, Power BM. Benefits and risks of furosemide in acute kidney injury.
Anaesthesia 2010;65:283-93.
3. Rejnmark L, Vestergaard P, Pedersen AR, Heickendorff L, Andreasen F,
Mosekilde L. Dose-effect relations of loop- and thiazide-diuretics on
calcium homeostasis: a randomized, double-blinded Latin-square mul-
tiple cross-over study in postmenopausal osteopenic women. Eur J Clin
Invest 2003;33:41-50.
4. Kubota T, Namba N, Kurotobi S, Kogaki S, Hirai H, Kitaoka T, et al. Ben-
eficial effect of oral bisphosphonate treatment on bone loss induced by
chronic administration of furosemide without alteration of its adminis-
tration and urinary calcium loss. Clin Pediatr Endocrinol 2006;15:101-
7.
5. Lim LS, Fink HA, Kuskowski MA, Taylor BC, Schousboe JT, Ensrud KE,
et al. Loop diuretic use and increased rates of hip bone loss in older men:
the Osteoporotic Fractures in Men Study. Arch Intern Med 2008;168:735-
40.
6. Sella S, Cattelan C, Realdi G, Giannini S. Bone disease in primary
hypercalciuria. Clin Cases Miner Bone Metab 2008;5:118-26.
7. Rejnmark L, Vestergaard P, Mosekilde L. Fracture risk in patients treated
with loop diuretics. J Intern Med 2006;259:117-24.
8. Xiao F, Qu X, Zhai Z, Jiang C, Li H, Liu X, et al. Association between loop
diuretic use and fracture risk. Osteoporos Int 2015;26:775-84.
9. Prandota J. Clinical pharmacology of furosemide in children: a supple-
ment. Am J Ther 2001;8:275-89.
10. Cheng HH, Carmona F, McDavitt E, Wigmore D, Perez-Rossello JM,
Gordon CM, et al. Fractures related to metabolic bone disease in chil-
dren with congenital heart disease. Congenit Heart Dis 2015;11:80-6.
11. Atkinson SA, Shah JK, McGee C, Steele BT. Mineral excretion in prema-
ture infants receiving various diuretic therapies. J Pediatr 1988;113:540-
5.
12. Texas Medicaid Enrollment Statistics. Texas Health and Human Service
Commission: Texas Health and Human Service Commission; 2015.
13. Karve S, Cleves MA, Helm M, Hudson TJ, West DS, Martin BC. Good
and poor adherence: optimal cut-point for adherence measures using ad-
ministrative claims data. Curr Med Res Opin 2009;25:2303-10.
14. Richardson P, Hawkey CJ, Stack WA. Proton pump inhibitors. Drugs
1998;56:307-35.
15. Fraser L-A, Leslie WD, Targownik LE, Papaioannou A, Adachi JD. The
effect of proton pump inhibitors on fracture risk: report from the Ca-
nadian Multicenter Osteoporosis Study. Osteoporos Int 2013;24:1161-
8.
16. FDA Drug Safety Communication. Possible increased risk of fractures of
the hip, wrist, and spine with the use of proton pump inhibitors. Atlanta
(GA): US Food and Drug Administration; 2011.
17. Austin PC. An introduction to propensity score methods for reducing the
effects of confounding in observational studies. Multivariate Behav Res
2011;46:399-424.
18. Arampatzis S, Gaetcke LM, Funk GC, Schwarz C, Mohaupt M,
Zimmermann H, et al. Diuretic-induced hyponatremia and osteopo-
rotic fractures in patients admitted to the emergency department. Maturitas
2013;75:81-6.
19. Heidrich FE, Stergachis A, Gross KM. Diuretic drug use and the risk for
hip fracture. Ann Intern Med 1991;115:1-6.
20. Campfield T, Braden G, Flynn-Valone P, Powell S. Effect of diuretics on
urinary oxalate, calcium, and sodium excretion in very low birth weight
infants. Pediatrics 1997;99:814-8.
21. Cheng HH, Carmona F, McDavitt E, Wigmore D, Perez-Rossello JM,
Gordon CM, et al. Fractures related to metabolic bone disease in chil-
dren with congenital heart disease. Congenit Heart Dis 2016;11:80-6.
Increased Fracture Risk with Furosemide Use in Children with Congenital Heart Disease
FLA 5.5.0 DTD ■ YMPD9926_proof ■ May 9, 2018
ORIGINAL ARTICLES
22. Hein G, Richter D, Manz F, Weitzel D, Kalhoff H. Development of neph-
rocalcinosis in very low birth weight infants. Pediatr Nephrol 2004;19:616-
20.
23. Knoll S, Alon US. Effect of thiazide on established furosemide-induced
nephrocalcinosis in the young rat. Pediatr Nephrol 2000;14:32-5.
24. Viswanathan S, Khasawneh W, McNelis K, Dykstra C, Amstadt R, Super
DM, et al. Metabolic bone disease: a continued challenge in extremely low
birth weight infants. JPEN J Parenter Enteral Nutr 2014;38:982-90.
25. Moffett BS, Price JF. National prescribing trends for heart failure medi-
cations in children. Congenit Heart Dis 2015;10:78-85.
26. Izumi G, Inai K, Shimada E, Nakanishi T. Vitamin D kinetics and para-
thyroid gland function in patients with congenital heart disease. Congenit
Heart Dis 2016;11:700-6.
27. Abou Zahr R, Faustino EV, Carpenter T, Kirshbom P, Hall EK, Fahey JT,
et al. Vitamin D status after cardiopulmonary bypass in children with con-
genital heart disease. J Intensive Care Med 2016;32:508-13.
28. Avitabile CM, Goldberg DJ, Zemel BS, Brodsky JL, Dodds K, Hayden-
Rush C, et al. Deficits in bone density and structure in children and young
adults following Fontan palliation. Bone 2015;77:12-6.
29. Shedeed SA. Vitamin D supplementation in infants with chronic con-
gestive heart failure. Pediatr Cardiol 2012;33:713-9.
30. Pilz S, Tomaschitz A, Drechsler C, Dekker JM, Marz W. Vitamin D
deficiency and myocardial diseases. Mol Nutr Food Res 2010;54:1103-
13.
31. Schlecht SH, Bigelow EM, Jepsen KJ. How does bone strength compare
across sex, site, and ethnicity? Clin Orthop Relat Res 2015;473:2540-7.
32. Adams AL, Shi JM, Reynolds K, Haque R, Cheetham TC, Kawatkar AA,
et al. Statins and hip fracture risk in men: a population-based case-
control study. Ann Epidemiol 2015;25:844-8.
33. Jain RK, Narang DK, Hans D, Vokes TJ. Ethnic differences in trabecular
bone score. J Clin Densitom 2016;20:172-9.
34. Pressley JC, Kendig TD, Frencher SK, Barlow B, Quitel L, Waqar F. Epi-
demiology of bone fracture across the age span in blacks and whites. J
Trauma 2011;71:S541-8.
35. Cauley JA. Defining ethnic and racial differences in osteoporosis and fra-
gility fractures. Clin Orthop Relat Res 2011;469:1891-9.
36. Freedberg DE, Haynes K, Denburg MR, Zemel BS, Leonard MB, Abrams
JA, et al. Use of proton pump inhibitors is associated with fractures in
young adults: a population-based study. Osteoporos Int 2015;26:2501-
7.
37. Sugiyama T, Torio T, Miyajima T, Kim YT, Oda H. Calcium, proton pump
inhibitors, and fracture risk. Osteoporos Int 2016;27:349-50.
38. Zhou B, Huang Y, Li H, Sun W, Liu J. Proton-pump inhibitors and risk
of fractures: an update meta-analysis. Osteoporos Int 2016;27:339-47.
39. Ghelani SJ, Spurney CF, Martin GR, Cross RR. Impact of pharmaco-
therapy on interstage mortality and weight gain in children with single
ventricle. Congenit Heart Dis 2013;8:219-27.
40. Moffett BS, Mattamal R, Ocampo EC, Petit CJ. Impact of pharmaco-
therapy on interstage outcomes in single ventricle infants. Congenit Heart
Dis 2011;6:286-93.
41. Slaughter JL, Stenger MR, Reagan PB, Jadcherla SR. Neonatal histamine-2
receptor antagonist and proton pump inhibitor treatment at United States
children’s hospitals. J Pediatr 2016;174:63-70, e3.
42. Albugeaey M, Alfaraj N, Garb J, Seiler A, Lagu T. Do hospitalists overuse
proton pump inhibitors? Data from a contemporary cohort. J Hosp Med
2014;9:731-3.
43. Eid SM, Boueiz A, Paranji S, Mativo C, Landis R, Abougergi MS. Pat-
terns and predictors of proton pump inhibitor overuse among aca-
demic and non-academic hospitalists. Intern Med 2010;49:2561-8.
44. McNally JD, Menon K. Vitamin D deficiency in surgical congenital heart
disease: prevalence and relevance. Transl Pediatr 2013;2:99-111.
45. Schoofs MW, van der Klift M, Hofman A, de Laet CE, Herings RM, Stijnen
T, et al. Thiazide diuretics and the risk for hip fracture. Ann Intern Med
2003;139:476-82.
7
THE JOURNAL OF PEDIATRICS • www.jpeds.com Volume ■■

Appendix

a. Index date: the date of the first prescription claim for furosemide or a random index date within 1 year from the first pre-
scription claim date.
b. Fracture: the date of the first diagnosis claims for fractures after index date or a random index date within 1 year from the
index date.

Appendix I. Study Timeframe.

7.e1 Heo et al

FLA 5.5.0 DTD ■ YMPD9926_proof ■ May 9, 2018

■■ 2018 ORIGINAL ARTICLES

Appendix II. ICD-9-CM Code for Congenital Heart Disease, Cardiomyopathy, and Heart Failure
ICD-9-CM Codes Description
Congenital heart disease
745.0 Common truncus
745.10 Complete transposition of great vessels
745.11 Double outlet right ventricle
745.12 Corrected transposition of great vessels
745.19 Other transposition of great vessels
745.2 Tetralogy of Fallot
745.3 Common ventricle
745.4 Ventricular septal defect
745.5 Ostium secundum type atrial septal defect
745.60 Endocardial cushion defect, unspecified type
745.61 Ostium primum defect
745.69 Other endocardial cushion defects
745.7 Cor biloculare
745.8 Other bulbus cordis anomalies and anomalies of cardiac septal closure
745.9 Unspecified defect of septal closure
746.00 Congenital pulmonary valve anomaly, unspecified
746.01 Atresia of pulmonary valve, congenital
746.02 Stenosis of pulmonary valve, congenital
746.09 Other congenital anomalies of pulmonary valve
746.1 Tricuspid atresia and stenosis, congenital
746.2 Ebstein's anomaly
746.3 Congenital stenosis of aortic valve
746.4 Congenital insufficiency of aortic valve
746.5 Congenital mitral stenosis
746.6 Congenital mitral insufficiency
746.7 Hypoplastic left heart syndrome
746.81 Subaortic stenosis
746.82 Cor triatriatum
746.83 Infundibular pulmonic stenosis
746.84 Obstructive anomalies of heart, not elsewhere classified
746.85 Coronary artery anomaly
746.86 Congenital heart block
746.87 Malposition of heart and cardiac apex
746.89 Other specified congenital anomalies of heart
746.9 Unspecified congenital anomaly of heart
747.0 Patent ductus arteriosus
747.10 Coarctation of aorta (preductal) (postductal)
747.11 Interruption of aortic arch
747.20 Anomaly of aorta, unspecified
747.21 Anomalies of aortic arch
747.22 Atresia and stenosis of aorta
747.29 Other anomalies of aorta
747.31 Pulmonary artery coarctation and atresia
747.32 Pulmonary arteriovenous malformation
747.39 Other anomalies of pulmonary artery and pulmonary circulation
747.40 Anomaly of great veins, unspecified
747.41 Total anomalous pulmonary venous connection
747.42 Partial anomalous pulmonary venous connection
747.49 Other anomalies of great veins
V15.1 Personal history of surgery to heart and great vessels, presenting hazards to health
Cardiomyopathy
425.0 Endomyocardial fibrosis
425.11 Hypertrophic obstructive cardiomyopathy
425.18 Other hypertrophic cardiomyopathy
425.2 Obscure cardiomyopathy of Africa
425.3 Endocardial fibroelastosis
425.4 Other primary cardiomyopathies
425.5 Alcoholic cardiomyopathy
425.7 Nutritional and metabolic cardiomyopathy
425.8 Cardiomyopathy in other diseases classified elsewhere
425.9 Secondary cardiomyopathy, unspecified
Heart failure
428.0 Congestive heart failure, unspecified
428.1 Left heart failure
428.20 Systolic heart failure, unspecified
428.21 Acute systolic heart failure
428.22 Chronic systolic heart failure
428.23 Acute on chronic systolic heart failure
428.30 Diastolic heart failure, unspecified
428.31 Acute diastolic heart failure
428.32 Chronic diastolic heart failure
428.33 Acute on chronic diastolic heart failure
428.40 Combined systolic and diastolic heart failure, unspecified
428.41 Acute combined systolic and diastolic heart failure
428.42 Chronic combined systolic and diastolic heart failure
428.43 Acute on chronic combined systolic and diastolic heart failure
428.9 Heart failure, unspecified

Increased Fracture Risk with Furosemide Use in Children with Congenital Heart Disease 7.e2

FLA 5.5.0 DTD ■ YMPD9926_proof ■ May 9, 2018

THE JOURNAL OF PEDIATRICS • www.jpeds.com Volume ■■

Appendix III. ICD-9-CM Code for Fractures, Bronchopulmonary Dysplasia, and Low Birth Weight or Prematurity
ICD-9-CM Codes Description
Fractures
733 Other disorders of bone and cartilage
767.2 Fracture of clavicle due to birth trauma
767.3 Other injuries to skeleton due to birth trauma
800 Fracture of vault of skull
801 Fracture of base of skull
802 Fracture of face bones
803 Other and unqualified skull fractures
804 Multiple fractures involving skull or face with other bones
805 Fracture of vertebral column without mention of spinal cord injury
806 Fracture of vertebral column with spinal cord injury
807 Fracture of rib(s), sternum, larynx, and trachea
808 Fracture of pelvis
809 Ill-defined fractures of bones of trunk
810 Fracture of clavicle
811 Fracture of scapula
812 Fracture of humerus
813 Fracture of radius and ulna
814 Fracture of carpal bone(s)
815 Fracture of metacarpal bone(s)
816 Fracture of one or more phalanges of hand
817 Multiple fractures of hand bones
818 Ill-defined fractures of upper limb
819 Multiple fractures involving both upper limbs, and upper limb with rib(s) and sternum
820 Fracture of neck of femur
821 Fracture of other and unspecified parts of femur
822 Fracture of patella
823 Fracture of tibia and fibula
824 Fracture of ankle
825 Fracture of one or more tarsal and metatarsal bones
826 Fracture of one or more phalanges of foot
827 Other, multiple, and ill-defined fractures of lower limb
828 Multiple fractures involving both lower limbs, lower with upper limb, and lower limb(s) with rib(s)and sternum
829 Fracture of unspecified bones
E887 Fracture, cause unspecified
V13.51 Pathologic fracture, Healed pathologic fracture
V13.52 Stress fracture
V15.51 Traumatic fracture, Healed traumatic fracture
V54.1 Aftercare for healing traumatic fracture
V54.2 Aftercare for healing pathologic fracture
V54.89 Aftercare for healing fracture NOS
V66.4 Following treatment of fracture
V67.4 Following treatment of healed fracture
Bronchopulmonary dysplasia
770.7 Chronic respiratory disease arising in the perinatal period
Low birth weight/prematurity
764.0 Light-for-dates infant without mention of fetal malnutrition
764.1 Light-for-dates infant with signs of fetal malnutrition
764.2 Fetal malnutrition without mention of light-for-dates
764.9 Fetal growth retardation unspecified
765.0 Disorders relating to extreme immaturity of infant
765.1 Disorders relating to other preterm infants
765.2 Weeks of gestation

7.e3 Heo et al

FLA 5.5.0 DTD ■ YMPD9926_proof ■ May 9, 2018

■■ 2018 ORIGINAL ARTICLES

a. Index date: the day when patients had first claims of diuretics.
b. Propensity score matching using a greedy algorithm with nearest available pair matching.

Appendix IV. Patient Attrition in the Texas Medicaid Database.

Increased Fracture Risk with Furosemide Use in Children with Congenital Heart Disease 7.e4

FLA 5.5.0 DTD ■ YMPD9926_proof ■ May 9, 2018

THE JOURNAL OF PEDIATRICS • www.jpeds.com Volume ■■

Appendix V. Site of Fractures for Original Population

Furosemide Furosemide No
adherent nonadherent furosemide
n % n % n % Total
Osteoporosis fracture 107 26.5 168 10.1 4418 28.2 4693
Fracture of radius and ulna 100 24.8 385 23.1 2552 16.3 3037
Fracture of humerus 19 4.7 250 15.0 1955 12.5 2224
Fracture of other and unspecified parts of femur 33 8.2 262 15.7 976 6.2 1271
Fracture of tibia and fibula 47 11.6 43 2.6 1111 7.1 1201
Fracture of >1 phalanges of hand 4 1.0 34 2.0 770 4.9 808
Fracture of clavicle — 0.0 62 3.7 513 3.3 575
Fracture of vault of skull 11 2.7 61 3.7 408 2.6 480
Fracture of ankle 5 1.2 26 1.6 415 2.7 446
Fracture of one or more tarsal and metatarsal bones — 0.0 45 2.7 317 2.0 362
Fracture of face bones 4 1.0 20 1.2 257 1.6 281
Fracture of rib(s), sternum, larynx, and trachea 1 0.2 30 1.8 227 1.4 258
Fracture of neck of femur — 0.0 83 5.0 174 1.1 257
Fracture of base of skull 30 7.4 35 2.1 156 1.0 221
Fracture of clavicle 1 0.2 6 0.4 197 1.3 204
Other and unqualified skull fractures 16 4.0 17 1.0 146 0.9 179
Fracture of pelvis — 0.0 — 0.0 177 1.1 177
Fracture, cause unspecified 4 1.0 16 1.0 126 0.8 146
Fracture of metacarpal bone(s) — 0.0 28 1.7 101 0.6 129
Fracture of unspecified bones 1 0.2 13 0.8 110 0.7 124
Fracture of >1 phalanges of foot — 0.0 6 0.4 116 0.7 122
Fracture of vertebral column without mention of spinal cord injury 15 3.7 — 0.0 99 0.6 114
Traumatic fracture — 0.0 13 0.8 78 0.5 91
Fracture of carpal bone(s) — 0.0 14 0.8 68 0.4 82
Ill-defined fractures of upper limb — 0.0 15 0.9 42 0.3 57
Following treatment of healed fracture — 0.0 — 0.0 42 0.3 42
Other, multiple, and ill-defined fractures of lower limb 6 1.5 — 0.0 34 0.2 40
Multiple fractures involving both lower limbs, lower with upper limb, — 0.0 32 1.9 8 0.1 40
and lower limb(s) with rib(s) and sternum
Fracture of scapula — 0.0 — 0.0 28 0.2 28
Pathologic fracture — 0.0 1 0.1 11 0.1 12
Fracture of vertebral column with spinal cord injury — 0.0 — 0.0 11 0.1 11
Fracture of patella — 0.0 — 0.0 8 0.1 8
Multiple fractures involving skull or face with other bones — 0.0 — 0.0 6 0.0 6
Ill-defined fractures of bones of trunk — 0.0 — 0.0 2 0.0 2
Multiple fractures of hand bones — 0.0 1 0.1 — 0.0 1
Total 404 100.0 1666 100.0 15 659 100.0 17 729

7.e5 Heo et al

FLA 5.5.0 DTD ■ YMPD9926_proof ■ May 9, 2018

■■ 2018 ORIGINAL ARTICLES

Appendix VI. Site of Fractures for Matched Population

Furosemide Furosemide No
adherent nonadherent furosemide
n % n % n % Total
Fracture of radius and ulna 82 35.7 182 31.2 251 17.6 515
Osteoporosis fracture 69 30.0 51 8.7 363 25.5 483
Fracture of other and unspecified parts of femur 5 2.2 46 7.9 187 13.1 238
Fracture of humerus 2 0.9 50 8.6 181 12.7 233
Fracture of neck of femur — 0.0 77 13.2 56 3.9 133
Fracture of vault of skull 9 3.9 32 5.5 46 3.2 87
Fracture of tibia and fibula 19 8.3 14 2.4 48 3.4 81
Fracture of ankle 5 2.2 8 1.4 60 4.2 73
Fracture of clavicle — 0.0 29 5.0 43 3.0 72
Fracture of >1 tarsal and metatarsal bones — 0.0 14 2.4 29 2.0 43
Other and unqualified skull fractures 16 7.0 7 1.2 14 1.0 37
Fracture of metacarpal bone(s) — 0.0 27 4.6 7 0.5 34
Fracture of pelvis — 0.0 — 0.0 33 2.3 33
Fracture of >1 phalanges of hand 1 0.4 2 0.3 24 1.7 27
Fracture of vertebral column without mention of spinal cord injury 15 6.5 — 0.0 11 0.8 26
Fracture of rib(s), sternum, larynx, and trachea — 0.0 5 0.9 21 1.5 26
Fracture of base of skull — 0.0 5 0.9 13 0.9 18
Fracture of unspecified bones 1 0.4 9 1.5 8 0.6 18
Fracture of one or more phalanges of foot — 0.0 6 1.0 8 0.6 14
Fracture, cause unspecified 4 1.7 4 0.7 5 0.4 13
Fracture of face bones — 0.0 3 0.5 7 0.5 10
Traumatic fracture — 0.0 6 1.0 4 0.3 10
Fracture of carpal bone(s) — 0.0 2 0.3 3 0.2 5
Ill-defined fractures of upper limb — 0.0 1 0.2 3 0.2 4
Fracture of clavicle — 0.0 2 0.3 — 0.0 2
Other, multiple, and ill-defined fractures of lower limb 2 0.9 — 0.0 — 0.0 2
Pathologic fracture — 0.0 1 0.2 — 0.0 1
Following treatment of healed fracture — 0.0 — 0.0 1 0.1 1
Total 230 100.0 583 100.0 1426 100.0 2239

Increased Fracture Risk with Furosemide Use in Children with Congenital Heart Disease 7.e6

FLA 5.5.0 DTD ■ YMPD9926_proof ■ May 9, 2018

THE JOURNAL OF PEDIATRICS • www.jpeds.com Volume ■■

Appendix VII. Sensitivity Analysis: Logistic Regression of Fractures by Furosemide Use Using an MPR Cutoff of 80%*
Categories OR 95% CI Wald Statistic P value†
Furosemide use‡
No furosemide Reference
Furosemide nonadherent 1.332 0.930-1.909 0.287 .12
Furosemide adherent 1.788 1.073-2.980 0.581 .03
Age 1.183 1.105-1.266 0.168 <.001
Sex
Male Reference
Female 0.880 0.654-1.184 −0.128 .40
Race/ethnicity
White Reference
Black 0.915 0.427-1.960 −.089 .82
Hispanic 0.892 0.551-1.443 −0.115 .64
Others 0.739 0.431-1.268 −0.302 .27
Only CHD§
No Reference
Yes 0.646 0.465-0.898 −0.437 .01
Bronchopulmonary dysplasia
No Reference
Yes 1.211 0.801-1.829 0.191 .36
Low birth weight/premature
No Reference
Yes 1.078 0.734-1.582 .075 .70
PPI
No Reference
Yes 1.875 1.314-2.676 0.629 .001
Heart-related surgery
No Reference
Yes 1.286 0.773-2.141 0.252 .33

*Likelihood ratio c2, 64.03 (P < .001); Model fit statistics (Akaike information criterion), 1498.50; percent concordant, 65.5%.
†P < .05.
‡Nonadherent furosemide ≤70% MPR first year after index; Adherent furosemide ≥70% MPR.
§Yes (only CHD) vs no (dual diagnosis CHD/cardiomyopathy or CHD/cardiomyopathy or only cardiomyopathy or only heart failure or comorbidity).

7.e7 Heo et al

FLA 5.5.0 DTD ■ YMPD9926_proof ■ May 9, 2018

■■ 2018 ORIGINAL ARTICLES

Appendix VIII. Sensitivity Analysis: Cox Proportional Hazard Model of Fractures by Furosemide Use Using an MPR
Cutoff of 80%*
Categories Hazard Ratio 95% CI* Coefficient P value†
Furosemide use‡
No furosemide Reference
Furosemide nonadherent 1.220 0.865-1.722 0.199 .26
Furosemide adherent 1.533 0.947-2.482 0.427 .08
Age 1.150 1.077-1.228 0.140 <.001
Sex
Male Reference
Female 0.882 0.664-1.172 −0.125 .39
Race/ethnicity
White Reference
Black 0.869 0.419-1.802 −0.141 .71
Hispanic 0.866 0.547-1.370 −0.144 .54
Others 0.781 0.467-1.306 −0.247 .35
Only CHD§
No Reference
Yes 0.692 0.506-0.946 −0.368 .02
Bronchopulmonary dysplasia
No Reference
Yes 1.175 0.794-1.739 0.161 .42
Low birth weight/premature
No Reference
Yes 1.087 0.753-1.570 .084 .65
PPI
No Reference
Heart-related surgery
No Reference
Yes 1.206 0.742-1.959 0.187 .45

*Likelihood ratio c , 46.55 (P < .001); model fit statistics (Akaike information criterion), 3107.52.
2

†P < .05.
‡Nonadherent furosemide ≤70% MPR first year after index; adherent furosemide ≥70% MPR.
§Yes (Only CHD) vs No (dual diagnosis CHD/cardiomyopathy or CHD/cardiomyopathy or only cardiomyopathy or only heart failure or comorbidity).

Increased Fracture Risk with Furosemide Use in Children with Congenital Heart Disease 7.e8

FLA 5.5.0 DTD ■ YMPD9926_proof ■ May 9, 2018

THE JOURNAL OF PEDIATRICS • www.jpeds.com Volume ■■

Appendix IX. Sensitivity Analysis: Logistic Regression of Fractures by Furosemide Use Controlling for H2-antagonist,
Beta-blocker, Calcium, and Vitamin D Supplement*
Category OR 95% CI Wald Statistic P value†
Furosemide use‡
No furosemide Reference
Furosemide nonadherent 1.456 1.043-2.034 0.376 .03
Furosemide adherent 1.660 1.032-2.670 0.507 .04
Age 1.150 1.075-1.229 0.140 <.001
Sex
Male Reference
Female 0.926 0.701-1.224 −0.077 .59
Race/ethnicity
White Reference
Black 0.487 0.239-0.995 −0.719 .05
Hispanic 0.625 0.414-0.941 −0.471 .03
Others 0.582 0.371-0.914 −0.541 .02
Only CHD§
No Reference
Yes 0.633 0.467-0.858 −0.457 .003
Bronchopulmonary dysplasia
No Reference
Yes 1.265 0.851-1.881 0.235 .25
Low birth weight/premature
No Reference
Yes 1.195 0.846-1.689 0.178 .31
PPI
No Reference
Yes 1.482 1.065-2.064 0.394 .02
Heart-related surgery
No Reference
Yes 1.123 0.663-1.901 0.116 .67
H2-antanogist
No Reference
Yes 1.178 0.830-1.671 0.164 .36
Beta-blocker
No Reference
Yes 1.642 0.969-2.782 0.496 .07
Calcium or vitamin D
No Reference
Yes 2.389 1.314-4.343 0.871 .004

*Likelihood ratio c2, 68.97 (P < .001); model fit statistics (Akaike information criterion), 1712.67; percent concordant, 64.8%.
†Significant at P < .05.
‡Nonadherent furosemide ≤70% MPR first year after index; adherent furosemide ≥70% MPR.
§Yes (only CHD) vs No (dual diagnosis CHD/cardiomyopathy or CHD/cardiomyopathy or only cardiomyopathy or only heart failure or comorbidity).

7.e9 Heo et al

FLA 5.5.0 DTD ■ YMPD9926_proof ■ May 9, 2018

■■ 2018 ORIGINAL ARTICLES

Appendix X. Sensitivity Analysis: Cox Proportional Hazard Model of Fractures by Furosemide Use Controlling for H2-
antagonist, Beta-Blocker, Calcium, and Vitamin D Supplement*
Categories Hazard Ratio 95% CI* Coefficient P value†
Furosemide use‡
No furosemide Reference
Furosemide nonadherent 1.294 0.939-1.783 0.258 .11
Furosemide adherent 1.379 0.88-2.161 0.322 .16
Age 1.120 1.05-1.195 0.113 .001
Sex
Male Reference
Female 0.933 0.714-1.219 −0.069 .61
Race/ethnicity
White Reference
Black 0.509 0.257-1.009 −0.675 .05
Hispanic 0.634 0.43-0.935 −0.456 .02
Others 0.665 0.433-1.02 −0.409 .06
Only CHD§
No Reference
Yes 0.686 0.513-0.918 −0.376 .01
Bronchopulmonary dysplasia
No Reference
Yes 1.263 0.868-1.84 0.234 .22
Low birth weight/premature
No Reference
Yes 1.206 0.864-1.683 0.187 .27
PPI
No Reference
Yes 1.369 0.999-1.875 0.314 .05
Heart-related surgery
No Reference
Yes 1.049 0.635-1.732 0.048 .85
H2-antanogist
No Reference
Yes 1.198 0.859-1.671 0.181 .29
Beta-blocker
No Reference
Yes 1.679 1.027-2.745 0.518 .04
Calcium or vitamin D
No Reference
Yes 2.338 1.365-4.005 0.849 .002

*Likelihood Ratio c2: 52.05 (P < .001); Model Fit Statistics (Akaike information criterion): 3,566.42.
†P < .05.
‡Nonadherent furosemide ≤70% MPR first year after index; Adherent furosemide ≥70% MPR.
§Yes (only CHD) vs no (dual diagnosis CHD/cardiomyopathy or CHD/cardiomyopathy or only cardiomyopathy or only heart failure or comorbidity).

Increased Fracture Risk with Furosemide Use in Children with Congenital Heart Disease 7.e10

FLA 5.5.0 DTD ■ YMPD9926_proof ■ May 9, 2018